23MAY2022 SNA/SNFO BOARD

[Shheen]

Did he email it to you through your command email or personal email? Sorry for all the questions, I’m in the dark on this and my CO & CCC aren’t updating me on anything. Plus, I’m seeing a lot of AD guys getting their finsel letter so I’m getting a little worried

My chain didn’t understand that they were required to reply with a release date so I had to schedule a meeting to get things moving along.

 

[elijahh36]

My chain didn’t understand that they were required to reply with a release date so I had to schedule a meeting to get things moving along.

How did you find out that your chain didn't reply with a release date? ( Did Mr Paul email you about your CO not giving a release date?). I ask this because I don't want to assume that my CO is holding up the process by not replying with a simply release date. Also how would you recommend I go about setting up a meeting to move things along? ( Assuming my CO didn't give a release date)

 

[Triumph_MAC]

How did you find out that your chain didn't reply with a release date? ( Did Mr Paul email you about your CO not giving a release date?). I ask this because I don't want to assume that my CO is holding up the process by not replying with a simply release date. Also how would you recommend I go about setting up a meeting to move things along? ( Assuming my CO didn't give a release date)

I would say respectfully asking him. However, I’m coming from a command that I had regular conversations with my CO both professional and just shooting the breeze.

 

[Maze_soba]

Hi all, I got bored over the weekend, so I finished the analysis for the May pilot board using the ML model I built, and there’s a decent amount to unpack.

First, I updated the training data to include results from the beginning of 2020, and quickly retrained the model.
Running the May board scores through, the model accurately predicted True 72% of the time, and False at 28%.

Here’s the prediction/actual matrix.

	Prediction
Actual	Prorec-Y	Prorec-N
Prorec-Y	33	4
Prorec-N	12	8

Taking a look at the false predictions:

• There are definitely weird exceptions. A 60 8/8/8/ and 67 8/9/8 were not selected. Maybe it was the age, but there must’ve been something else going on with the package, or this was bad data.
• Some people got hyper lucky getting in with a 5/5/5. By all estimates, this would have been a N based on the model, but they somehow got it. The “N” prediction also probably has to do with the fact that only the nerds post their scores here so the model is biased to higher scores.
• Long story short, the model is doing a pretty decent job right now.

Next, I added the May board results to the training data (sans ISPP) and it should be much more robust than the first version I made last December. If I get really bored before shipping out, I might see about making a simple web app.

A couple of things I’ve noticed after retraining + analysis of the new model:

• OAR and PFAR have the most impact on prorec status with each contributing about 24% to the final result. This isn’t to say the OAR is important to getting a Y, but that it might mean you’re more likely to do better on PFAR if you have a higher OAR. Correlation != causation and all that jazz.
• Age and GPA are the next most important, with each contributing 10% to the final result. Age obviously is a negative impact, but it isn’t a huge amount.
• Flight exp, prior service, and sex still have basically no impact on the outcome (combined, only contributes ~9% to the final result)

And to those of you who requested that I run through your scores, I’ll be PMing you the results!

*disclaimer: this analysis is pretty half-assed and there are big holes in it. But, if you still want me to run your scores though, please let me know.

 

[extraschmedium]

Hi all, I got bored over the weekend, so I finished the analysis for the May pilot board using the ML model I built, and there’s a decent amount to unpack.

First, I updated the training data to include results from the beginning of 2020, and quickly retrained the model.
Running the May board scores through, the model accurately predicted True 72% of the time, and False at 28%.

Here’s the prediction/actual matrix.

	Prediction
Actual	Prorec-Y	Prorec-N
Prorec-Y	33	4
Prorec-N	12	8

Taking a look at the false predictions:
View attachment 35927

• There are definitely weird exceptions. A 60 8/8/8/ and 67 8/9/8 were not selected. Maybe it was the age, but there must’ve been something else going on with the package, or this was bad data.
• Some people got hyper lucky getting in with a 5/5/5. By all estimates, this would have been a N based on the model, but they somehow got it. The “N” prediction also probably has to do with the fact that only the nerds post their scores here so the model is biased to higher scores.
• Long story short, the model is doing a pretty decent job right now.

Next, I added the May board results to the training data (sans ISPP) and it should be much more robust than the first version I made last December. If I get really bored before shipping out, I might see about making a simple web app.

A couple of things I’ve noticed after retraining + analysis of the new model:

• OAR and PFAR have the most impact on prorec status with each contributing about 24% to the final result. This isn’t to say the OAR is important to getting a Y, but that it might mean you’re more likely to do better on PFAR if you have a higher OAR. Correlation != causation and all that jazz.
• Age and GPA are the next most important, with each contributing 10% to the final result. Age obviously is a negative impact, but it isn’t a huge amount.
• Flight exp, prior service, and sex still have basically no impact on the outcome (combined, only contributes ~9% to the final result)

And to those of you who requested that I run through your scores, I’ll be PMing you the results!

*disclaimer: this analysis is pretty half-assed and there are big holes in it. But, if you still want me to run your scores though, please let me know.

My man, can you run my scores?

 

[exNavyOffRec]

Hi all, I got bored over the weekend, so I finished the analysis for the May pilot board using the ML model I built, and there’s a decent amount to unpack.

First, I updated the training data to include results from the beginning of 2020, and quickly retrained the model.
Running the May board scores through, the model accurately predicted True 72% of the time, and False at 28%.

Here’s the prediction/actual matrix.

	Prediction
Actual	Prorec-Y	Prorec-N
Prorec-Y	33	4
Prorec-N	12	8

Taking a look at the false predictions:
View attachment 35927

• There are definitely weird exceptions. A 60 8/8/8/ and 67 8/9/8 were not selected. Maybe it was the age, but there must’ve been something else going on with the package, or this was bad data.
• Some people got hyper lucky getting in with a 5/5/5. By all estimates, this would have been a N based on the model, but they somehow got it. The “N” prediction also probably has to do with the fact that only the nerds post their scores here so the model is biased to higher scores.
• Long story short, the model is doing a pretty decent job right now.

Next, I added the May board results to the training data (sans ISPP) and it should be much more robust than the first version I made last December. If I get really bored before shipping out, I might see about making a simple web app.

A couple of things I’ve noticed after retraining + analysis of the new model:

• OAR and PFAR have the most impact on prorec status with each contributing about 24% to the final result. This isn’t to say the OAR is important to getting a Y, but that it might mean you’re more likely to do better on PFAR if you have a higher OAR. Correlation != causation and all that jazz.
• Age and GPA are the next most important, with each contributing 10% to the final result. Age obviously is a negative impact, but it isn’t a huge amount.
• Flight exp, prior service, and sex still have basically no impact on the outcome (combined, only contributes ~9% to the final result)

And to those of you who requested that I run through your scores, I’ll be PMing you the results!

*disclaimer: this analysis is pretty half-assed and there are big holes in it. But, if you still want me to run your scores though, please let me know.

High PFAR and non select often are due to something negative in the application, but not always sometimes they just miss a good application.

5's being selected is rare, they were not ISPP right? and not AD?

The age thing might be an anomaly, if they meet the age requirement and can graduate OCS in time that is not supposed to be a positive or negative thing.

 

[vbts13]

Question for those who were selected SNA on this board... My recruiter called me this week and said he was instructed to ask those slated for FY24 if they're able leave on a moments notice... and if so they will be placed on a list and potentially sent sooner rather than later... Anyone else hear anything similar from their recruiter? I'm a bit skeptical because of the stockpile of SNAs waiting to class up in Pensacola... makes no sense why'd they rush more selects through OCS to hurry up and wait

 

[BDavis11]

Question for those who were selected SNA on this board... My recruiter called me this week and said he was instructed to ask those slated for FY24 if they're able leave on a moments notice... and if so they will be placed on a list and potentially sent sooner rather than later... Anyone else hear anything similar from their recruiter? I'm a bit skeptical because of the stockpile of SNAs waiting to class up in Pensacola... makes no sense why'd they rush more selects through OCS to hurry up and wait

Hurry up and wait is the name of the military game

 

[Maze_soba]

High PFAR and non select often are due to something negative in the application, but not always sometimes they just miss a good application.

5's being selected is rare, they were not ISPP right? and not AD?

The age thing might be an anomaly, if they meet the age requirement and can graduate OCS in time that is not supposed to be a positive or negative thing.

The triple 5's weren't ISPP, but were AD. I don't know if the individual just forgot to label as ISPP but it's definitely rare.

Age is surprisingly correlated to the results! Looking at the broader data, it seems like the older a person is, the higher scores they need to get in.

 

[Furnace01]

Hi all, I got bored over the weekend, so I finished the analysis for the May pilot board using the ML model I built, and there’s a decent amount to unpack.

First, I updated the training data to include results from the beginning of 2020, and quickly retrained the model.
Running the May board scores through, the model accurately predicted True 72% of the time, and False at 28%.

Here’s the prediction/actual matrix.

	Prediction
Actual	Prorec-Y	Prorec-N
Prorec-Y	33	4
Prorec-N	12	8

Taking a look at the false predictions:
View attachment 35927

• There are definitely weird exceptions. A 60 8/8/8/ and 67 8/9/8 were not selected. Maybe it was the age, but there must’ve been something else going on with the package, or this was bad data.
• Some people got hyper lucky getting in with a 5/5/5. By all estimates, this would have been a N based on the model, but they somehow got it. The “N” prediction also probably has to do with the fact that only the nerds post their scores here so the model is biased to higher scores.
• Long story short, the model is doing a pretty decent job right now.

Next, I added the May board results to the training data (sans ISPP) and it should be much more robust than the first version I made last December. If I get really bored before shipping out, I might see about making a simple web app.

A couple of things I’ve noticed after retraining + analysis of the new model:

• OAR and PFAR have the most impact on prorec status with each contributing about 24% to the final result. This isn’t to say the OAR is important to getting a Y, but that it might mean you’re more likely to do better on PFAR if you have a higher OAR. Correlation != causation and all that jazz.
• Age and GPA are the next most important, with each contributing 10% to the final result. Age obviously is a negative impact, but it isn’t a huge amount.
• Flight exp, prior service, and sex still have basically no impact on the outcome (combined, only contributes ~9% to the final result)

And to those of you who requested that I run through your scores, I’ll be PMing you the results!

*disclaimer: this analysis is pretty half-assed and there are big holes in it. But, if you still want me to run your scores though, please let me know.

Would you mind running my scores? I still think I'm going to retake, but I'm just curious

 

[Ed014]

Question for those who were selected SNA on this board... My recruiter called me this week and said he was instructed to ask those slated for FY24 if they're able leave on a moments notice... and if so they will be placed on a list and potentially sent sooner rather than later... Anyone else hear anything similar from their recruiter? I'm a bit skeptical because of the stockpile of SNAs waiting to class up in Pensacola... makes no sense why'd they rush more selects through OCS to hurry up and wait

I was selected SNFO but I have seen a handful of SNA's leave sooner than expected. From my OCS class group chat one of our guys who was an SNA was asked if he wanted to leave in week and he accepted it. I have seen some more be asked in other group chats as well. Someone said SNA's are dropping before leaving to OCS because of the potential long wait time in Pensacola. So from observing it seems like they are replacing the spots with other SNA's not adding more.

 

[smpl_dude]

I was selected SNFO but I have seen a handful of SNA's leave sooner than expected. From my OCS class group chat one of our guys who was an SNA was asked if he wanted to leave in week and he accepted it. I have seen some more be asked in other group chats as well. Someone said SNA's are dropping before leaving to OCS because of the potential long wait time in Pensacola. So from observing it seems like they are replacing the spots with other SNA's not adding more.

Imagine giving up your golden ticket into naval aviation because you might have to sit around in the Florida sun for a bit longer than normal.

 

[space_sailor]

Imagine giving up your golden ticket into naval aviation because you might have to sit around in the Florida sun for a bit longer than normal.

absolutely bonkers - while making a solid paycheck too that most aren't even near

 

[carviator]

Imagine giving up your golden ticket into naval aviation because you might have to sit around in the Florida sun for a bit longer than normal.

After being on sea duty for the past 5 years I'm looking forward to a little break in Pensacola..I wouldn't mind even a six month class up wait lol.

 

[exNavyOffRec]

The triple 5's weren't ISPP, but were AD. I don't know if the individual just forgot to label as ISPP but it's definitely rare.

Age is surprisingly correlated to the results! Looking at the broader data, it seems like the older a person is, the higher scores they need to get in.

As long as a person meets the age requirement in the PA without a waiver they are to be considered, to use age in selection criteria would be considered discrimination.